Ask A Banker: Capital, Capital!
Q. Should banks be required to hold much more capital as a safety net? Or should they be putting that money to productive use by lending it out instead?
Naaaaah, just messing with you. The recent debate over bank capital, sparked by Anat Admati and Martin Hellwig's book The Bankers' New Clothes and the Brown-Vitter bill in the Senate, has had at least one unambiguous effect, which is that it is now fashionable in certain circles to say "banks don't hold capital" in the tone of voice formerly reserved for correcting split infinitives or declining ketchup on your hot dog.
Banks don't hold capital! And bank capital isn't money kept in a vault - it can be put to productive use, too.
Okay, that's out of the way. Now, let's talk about what capital is and whether there should be more of it. I feel a little silly doing so because, while The Bankers' New Clothes is sort of the Eats, Shoots and Leaves of financial regulation, it's also a pretty good book about bank capital that is designed to be super easy for the non-expert reader, so you could just read it and I could go do something else.
Nonetheless! The book seems to have sparked the Brown-Vitter bill in the Senate, which would require big (over $500 billion in assets) banks to have capital equal to at least 15% of their assets. This in turn would require those big banks to raise something over $1 trillion in capital or, alternatively, to become much smaller banks.* The banks don't want to get smaller, for reasons you can probably figure out on your own; they also don't seem all that enamored of raising more capital, for reasons that might require more explaining.
So: What is capital? We might as well start with "banks don't hold capital": capital isn't a thing a bank has or holds or keeps in a vault somewhere collecting dust; it's more of a description of where the money comes from. Or, doesn't come from: capital is money that isn't borrowed. Banks, like all companies, own stuff, and like many companies, they borrow money. Here's a roughly accurate diagram of what that looks like at one particular bank (J.P. Morgan):**
Notice that the left side (JPMorgan's "assets") is bigger than the right side (its "liabilities"), which is generally the case - companies that owe more than they own tend to be in trouble. JPMorgan has about $2.4 trillion worth of stuff, some of which represents money put to productive use in the economy (loans to people to buy houses, or loans to companies to build factories), some of which is just money kept in vaults and emergency funds, and some of which is ... derivatives, where, you make the call.
JPMorgan also owes around $2.2 trillion to various people. Some of those people can take their money out any time - like people with Chase checking accounts (under "Deposits"), or people who lend to JPMorgan in the repo markets (under "Other short-term debt"). Others have their money locked in for a while - like people with 6-month certificates of deposit (also under "Deposits"), or people who own JPMorgan ten-year bonds ("Long-term debt").***
If you subtract what JPMorgan owes other people from what it has, you get its capital, which as it happens is about $207 billion:
So, on some simplistic assumptions, capital is just the result of some arithmetic:
Capital is how much money would be left in the bank if you sold all the bank's stuff and paid off all its debts.
If you actually did that - and you wouldn't, it wouldn't really work out this way, but pretend - if you actually did that then that leftover money would go to JPMorgan's shareholders. They're what's called the "residual claimants," which just means that they get what's left after everyone else has been paid off.**** So the capital is sometimes called "shareholders' capital," because it belongs to them.
Where does capital come from? Well, it comes from arithmetic doesn't it? At the birth of a bank, some people get together and decide to start a bank and they each contribute $100 for one share, or whatever, and that's the initial capital that gets the bank started. But over time, the bank's capital mostly grows from arithmetic: if the bank is making profits, then the value of what it owns will grow faster than the value of what it owes. In other words, if a bank is making a profit, its capital will increase.
For reasons we'll get to in a bit, banks sometimes have too much or too little capital - too much if they've been really profitable, too little if they've lost a lot of money. If they have too much capital they tend to return money to shareholders, by buying back stock or paying dividends. If they have too little they tend to raise more capital, by selling more shares. Another arithmetic definition of capital is that it's:
- the total amount of profit the bank has made in its history (minus, of course, all the losses it's made in its history), plus
- the amount of money it's raised by selling stock, minus
- the amount it's spent buying back stock and paying dividends.
The fact that "profit plus cash that comes in from selling stock minus cash returned to shareholders" comes out to the same number as "the value of your assets minus your liabilities" is the sort of mild arithmetico-definitional magic that really ought to have a name, like "fundamental theorem of accounting" or something, but as far as I can tell it doesn't.*****
Those arithmetic definitions of capital should make clear an important fact: if a bank loses money, its capital is reduced before any of its debts are. The capital is the "first-loss position"; sometimes people say it "absorbs" losses. If JPMorgan just misplaces $100 billion of cash, then its assets will go from $2.4 trillion to $2.3 trillion, but its debts won't change by a penny: it'll still owe various bondholders, depositors, etc. a total of about $2.3 trillion. This means that its capital will now be $107 billion, say, instead of $207 billion: the entire $100 billion loss will go directly to the people who own stock in JPMorgan — the people with a claim on JPMorgan's capital. If JPMorgan misplaces $300 billion of cash, then its capital will be zero, and the stock in JPMorgan will be worth zero. On top of that, some of the people who loaned money to JPMorgan won't be paid back.
Those arithmetic definitions might raise another question: if capital is, at the end of the day, the value of what you have minus what you owe to other people, or equivalently the value of all your profits added up, then why wouldn't you want more of it? The big banks are subject to capital requirements by regulation, and tend in practice not to have too much more capital than they're required to have. And they are very unhappy with Brown-Vitter, which would raise those requirements dramatically.
But, why? In your personal life having more money and less debt seems like a good thing, no? Most people celebrate paying off the mortgage, rather than running out and getting another one. Why do bank executives disagree?
The answer starts with the fact that bank executives, in theory, work for (and generally are) shareholders - the people with the claim on capital. Those shareholders obviously want the bank to increase its capital by making money. But then they want the bank to give them the money: to "return capital" via share repurchases (which drive up the stock price) and dividends(cash payments to shareholders). For the bank to make money and keep it is relatively less appealing, and what they particularly don't want is for the bank to increase its capital by asking them for money.******
I'm tempted to stop there and say it's as simple as that: people invest in banks because they'd like to make money; having money come out of the bank and into their pockets is good; having money stay in the bank is so-so; having the bank sell more shares is bad.
But there's another issue, which is that banks - and their shareholders - tend to like leverage, which is the superpower that borrowed money creates. Borrowing money - especially when you can borrow it really cheaply, like you can today - allows you to magnify your profits and losses. Magnifying your profits is good for shareholders (they get the profits!). Magnifying your losses isn't great, but since the shareholders don't necessarily suffer all of the losses (their shares can't go below zero), they might still prefer to take the risk. "Capital," remember, just refers to money that the bank hasn't borrowed: the more capital a bank has, by definition, the less leverage it has.
On the other hand, people who think that banks need to be safer like capital and tend to think that there should be more of it. Capital has two features that make it attractive to these people. Or, one feature, which is that you don't have to pay it back. But this means two things: first, you don't have to pay it back at any particular time. Second, there's no particular amount that you have to pay back.
First, time: much bank financing (tri-party repo, for instance) is what is called "overnight" financing because it is literally that: someone is lending you money for less than 24 hours. This is pretty annoying for all concerned, but it has the key advantage for the lender that they can get out at the first sign of trouble. If you've been lending money overnight to a bank for a while and find out on Wednesday morning that they've lost a lot of money to rogue traders, you can just decide not to "roll" your loan, that is, lend the money again Wednesday night. And, there, you're done, you've got your money back.
That's bad for the bank: when they've lost a lot of money to rogue traders is when they particularly need money, and might have a tough time raising it (because who wants to lend to a bank that might be going under?). Look back at that chart: JPMorgan has something like $650 billion of short-term debt - some overnight, but all of it coming due within one year. If they couldn't "roll" that debt then they'd need to pay it all back in fairly short order. They have about $300 billion in cash, which would help, but if they really needed to pay back the full $650 billion they'd need to sell $350 billion of other stuff to raise more cash. Selling $350 billion worth of stuff all at once has a tendency to make it worth less than $350 billion.
This problem - which is often referred to under the name "liquidity" - can be addressed by having lots of capital. Capital never needs to be paid back. Shareholders can never demand that JPMorgan pay them back their $207 billion, so that funding can never cause a crisis.
But the liquidity problem can also be addressed in other ways. For instance: banks could just use less short-term debt, which can cause a liquidity crisis any given day, and more long-term debt, which makes these crises less likely. And in fact there are a lot of proposals to require banks to use more long-term debt and rely less on short-term funding markets. Or: banks could have more liquid assets - more stuff that they could sell quickly, without it losing value, if they did need to pay off their debts suddenly - and, again, that's a popular topic of regulation.
The second advantage of capital is unique to capital:******* you don't have to pay back any particular amount. A bank share, unlike a bank bond or deposit, doesn't have a fixed dollar amount; it's just worth whatever it's worth. This sounds like a meaningless advantage because, as we just said, you never need to pay it back at all, but there's nonetheless some psychologico-regulatory meaning that's worth discussing. It goes like this:
- If you lend money to a bank, for example by putting your life savings in a checking account at the bank, and the bank loses all of it, you will freak out, and so will your congressperson; whereas
- If you invest money in bank stock, and the value of that stock goes to zero, it's really your fault, man. You can freak out if you want, and the world being what it is you can probably find someone to get angry along with you, but it ain't me. You bought bank stocks, you got what you deserve.
Is that ... unsatisfying? Because it's kind of important. The thing about "more capital makes banks safer" is that in a certain light it's nonsensical: a bank's risk of losing money depends on its actions and its assets. If it buys terrible mortgage-backed securities or gets sued for fraud all the time, then it will lose money, no matter how much money it owes or doesn't owe to other people. The question is just who will bear those losses.********
But that's an important question. If JPMorgan, with its $207 billion in capital, loses $50 billion, then to some approximation its shareholders will suffer $50 billion in losses. Tough luck to them! But if it loses $500 billion, then the shareholders' $207 billion of losses are the least of our problems. The big problem is where the other $300 billion in losses go: which of the people holding the $2.2 trillion in JPMorgan debt don't get paid back (or don't get fully paid back).
There are not a lot of great answers. The depositors? Like, JPMorgan's failure should cause you to lose what's in your checking account? Nah, we have deposit insurance because that seems like such a bad idea. The creditors and trading counterparties? Maybe, but remember that those tend to be financial institutions too; if they take unexpected losses on JPMorgan debts that they thought were totally safe, then they might run into trouble too, causing a domino effect of financial panic.
The answer is probably: there'll be a taxpayer bailout. Hahaha, no, the whole post-financial-crisis regulatory world is oriented around finding ways to avoid a bailout. There'll be an "orderly liquidation" where a carefully managed process causes some of a failed bank's creditors to take losses without sparking a panic or touching the creditors (depositors, etc.) whom we don't want touched. I mean, sure, whatever, maybe. But there's no doubt that the simplest way to reduce the risk of panic on the one side, and bailouts on the other, is to raise capital requirements in advance: to force banks to get relatively more of their money from people who expect to bear losses, and relatively less from people who don't.
The advantage of making banks have much more capital is that it makes the risks clearer: it forces banks to be funded more by shareholders who knowingly accept the risk that they might lose money, and less by depositors and repo markets who pretend that their financing of bank activities is risk-free.
The disadvantage of making banks have much more capital is, also, that it makes the risks clearer. Banks exist to hide risks: to take money entrusted to them for safekeeping and lend it out to fund the risky activities that help our economy grow, without scaring people about the risk. Banks "exist to hide such information; they are created and structured to keep secrets." The secret being: someone might lose money!
But sometimes that's a good secret to keep. Ultimately banks get their money from people. People seem to have some preference for putting money into banks in "safe" forms (deposits, etc.), rather than "risky" forms (shares of stock). If regulation shifts the mix into more "risky" forms of bank financing - if it makes people face up to the risks of their banks - then they will have a harder time satisfying that preference for safe assets. Then what happens?
Maybe we all just get wiser and more mature about our risks, and fund our risky economy with a clear-eyed, world-weary stoicism. But: when has that happened? Maybe we shadow-bank: we shift our money into other, non-bank, areas that promise us safety but carry their own risks of financial disaster. (And that are less blessed with enlightened regulation than the banking sector is.) Or maybe we just reduce the size of our banking sector, which a lot of people might like, but which might also mean less lending, less financing for new businesses, and less economic activity.
Which, when our economy is shaky to begin with, might turn out to be a high price to pay for facing the demons of our banking system. Increasing bank capital may force everyone to be more realistic about the risks of banking, but realistic isn't always what we want.
Correction: The following sentence originally said "billion" when it should have said "trillion." Thanks to the commenter who pointed out the error. If JPMorgan just misplaces $100 billion of cash, then its assets will go from $2.4 trillion to $2.3 trillion, but its debts won't change by a penny: it'll still owe various bondholders, depositors, etc. a total of about $2.3 trillion.
* Though, amusingly, the bill seems to prohibit that approach. The legislative process is not always pretty.
** This is not meant particularly seriously; you could carve up the categories differently, and probably a lot of what I'm calling "monstrosities" could be more accurately lumped with cash, or stocks and bonds, than with the core monstrosities of opaque derivatives.
*** There's also a tiny sliver of preferred stock. People who own JPMorgan preferred stock can never demand their money back - they just put up, say, $10,000, and get back, say, $550 a year in interest for the rest of eternity. Under certain definitions preferred stock is capital; under others it isn't. That's kind of a boring debate? Under U.S. accounting rules it's normally "equity" rather than "liabilities," so my presentation here is somewhat nonstandard, but I'm conforming to the Brown-Vitter approach, which believes that the only real capital is common equity capital.
**** There is some interesting controversy over whether the shareholders are really the "residual claimants," or at least, over whether they're the only ones. You might argue that if there's any money left over after JPMorgan's creditors have been paid back, JPMorgan's bankers and traders will take it, and you wouldn't necessarily be wrong.
***** Loosely speaking it's called "the accounting equation," but only very loosely speaking.
****** This is especially true because, these days, giving a bank a dollar makes it worth a bit less than a dollar. If you go to Google Finance you'll see that JPMorgan's stock - all of it combined, its "market capitalization" - is worth about $193 billion, not $207 billion. So the market places a lower value on JPMorgan's stuff than JPMorgan's accountants do. The market thinks JPMorgan is worth less than its accounting value ("book value") because the market thinks its future is somewhat alarming, or because the market distrusts the value that its accountants assign to its assets, or some combination of the two. What this means is that if JPMorgan raises $1 from shareholders, it will only be worth about 93 cents, and if JPMorgan gives $1 back to shareholders, it'll only cost them about 93 cents. I don't mean to pick on JPMorgan; most big banks these days trade at a discount to their book value due to mistrust and fears about the future. Bank of America's dollars are worth 67 cents.
******* Though not unique to common equity capital. Some other things, like preferred stock or "bail-in debt," is also loss absorbing. The Brown-Vitter and Admati-Hellwig approach stresses simplicity so is deeply distrustful of those sorts of things.
******** I'm lying. The question isn't just that. There are real costs of debt, which are lumped under the heading "costs of insolvency"; roughly speaking, if you lose enough money, then the holders of your debt can put you into bankruptcy and force you to sell all your stuff at a loss and so destroy value. Whereas capital can't do that: if a well capitalized bank loses a lot of money (but less than its capital cushion), it can live to fight another day. There is a lot to say about these issues, and Admati and Hellwig talk more about them in chapter 3 of their book.